Issue link: https://cp.revolio.com/i/661564
6 | THE DATA CENTER JOURNAL www.datacenterjournal.com to the consumer. In the hyperconvergence world, there is tangible value in having the building blocks created for any business- savvy end user, as the time it takes com- plete the final assembly and testing can't be discounted." Bhattacharjee offers a similar per- spective. "e bigger savings comes from the single management platform. Hyper- convergence, when designed appropriately, significantly reduces operational costs in addition to reducing total cost of owner- ship (TCO) even further." Much of this value is in the amortization of integration costs—including manufacturing infra- structure as well as expertise in this area— across multiple customers. e potential for savings on the capital side may grow, however. Harris describes the scenario as follows: "One of the factors to keep in mind with regard to cost is the primary use of open and white- box-style solutions. Although end users may naturally choose servers from their Tier One provider historically, when they purchase the building-block equivalent, that supplier usually starts with white-box components and then supports them. In essence, the hyperconverged suppliers have the luxury of using high-quality Tier Two components to build their Tier One converged building block, lowering the comparative costs." Does this hyPerConvergenCe Look gooD on me? When budgets are tight, who needs a Ferrari when a Toyota will do? e ques- tion is then whether hyperconvergence is a fit for a particular business model or facility. Considerations such as data center size, maturity and performance needs all factor into the answer. For instance, ow- ing to the modularity of this technology, data center size isn't necessarily an issue, excepting perhaps very large or very small facilities. Notes Harris, "Owing to the soware and high-performance-networking layers involved in hyperconvergence, there is no practical scale limitation for Fortune 500 data centers. Although some may argue that the largest web-scale properties should be cautious in their usage, even the largest enterprise end users can deploy hy- perconverged appliances without concern." A possible barrier for very small data cen- ters (perhaps stretching the definition of the term data center a bit) is the unit cost for implementing hyperconvergence. "e use of hyperconverged solutions forces capacity planning to be done at the macro level with the unit of one building block. In traditional data centers a single $10,000 server, switch or storage array could be added at will, but in a hyperconverged center the unit of measure is an entire rack, which may have a cost of $250,000 or more. Hyperconverged solutions can make it easier to deploy new applications, but there is a step function required to bring in new building blocks." ese numbers could be intimidating to a small operator, although in such cases, outsourced infra- structure may be the better option. Another concern is whether the ap- proach is amenable to an existing facility. For some technologies, the only way to avoid gratuitous expenses or technical challenges is to include these technologies in the original design rather than trying to cram them in through a retrofit. For oth- ers, adding them in later is less onerous. "Hyperconvergence works very well in small or large data center environments," said Bhattacharjee. "You can have a hyper- converged system exist side by side with a legacy system and expand the hyper- converged environment over time. at's the beauty of being able to start small and grow on the basis of need." Expanding on this sentiment, Har- ris said, "Hyperconverged solutions can be deployed in nearly any data center, supporting new or existing applications. e soware management tools and processes may change slightly, but there is no practical reason to limit the use of hyperconverged solutions." He notes, however, that owing to their relatively high density, the data center operator must take into account the weight/loading, power consumption and cooling needs of these solutions. Such factors can influence allowable placement, capacity and even feasibility in a given facility. ConCLusions Any hot new technology poses a danger in that the hype can easily draw in unsuspecting data center operators look- ing to ease their task or save some money. Reckless adoption can lead to higher costs, project delays and even downtime. But a relatively new or immature technology can still have value to those who carefully weigh the costs and benefits. Hypercon- vergence seeks to continue the trend of virtualization and modularity in the data center. Providing a good summary, Harris said, "From an operations point of view, the simplification of the computing func- tion in macro building blocks reduces the overhead and costs required to maintain hyperconverged solutions. Since there is more consistency in the building blocks, there are fewer opportunities for human error, as the building blocks all look and operate the same." ese features of hyper- convergence enable a modular approach to capacity: "When data centers need to scale, adding new building blocks is easy without fear of affecting anything that's already in place, allowing downtime to be reduced. e hyperconverged approach creates resources as services, which are consumable by any VM and application." Concerns such as vendor lock-in remain for hyperconvergence, but industry trends toward open source are countering this potential problem. Cost is some- thing of a mixed bag. Nevertheless, when considering any technology, the key is to first know yourself: identify what your data center needs and what your priorities are. Only then will you be able to weigh the ups and downs of hyperconvergence and make a decision that helps rather than hinders operations. n